{"id":655893,"date":"2020-05-29T12:21:47","date_gmt":"2020-05-29T19:21:47","guid":{"rendered":"https:\/\/www.microsoft.com\/en-us\/research\/?post_type=msr-event&p=655893"},"modified":"2025-08-06T11:52:48","modified_gmt":"2025-08-06T18:52:48","slug":"frontiers-in-machine-learning-2020","status":"publish","type":"msr-event","link":"https:\/\/www.microsoft.com\/en-us\/research\/event\/frontiers-in-machine-learning-2020\/","title":{"rendered":"Frontiers in Machine Learning 2020"},"content":{"rendered":"\n\n

Contact Us:<\/strong> If you have questions about this event, please email us at mlevent@microsoft.com<\/a><\/p>\n

\n\t\n\t\tWatch on-demand\t<\/a>\n\n\t<\/div>\n

Opens in a new tab<\/span><\/p>\n

The first virtual Frontiers in Machine Learning event took place from July 20-23, 2020.<\/p>\n

This four-day virtual conference brought together academics, researchers, and PhD Students. The program was rich, engaging, and filled with current themes and research outcomes spanning theory and practice in Machine Learning. The agenda covered talks and discussions with Microsoft researchers and academic collaborators.<\/p>\n

Agenda Overview<\/h3>\n\n\n\n\n\n\n\n\n\n\n\n
Date <\/strong><\/td>\n Time <\/strong><\/td>\n Program <\/strong><\/td>\n<\/tr>\n
Monday, July 20, 2020<\/td>\n9:00 AM\u201310:00 AM PDT<\/td>\nFireside Chat, Chris Bishop and Peter Lee<\/td>\n<\/tr>\n
<\/td>\n10:30 AM\u201312:00 PM PDT<\/td>\nMachine Learning Conversations, a panel led by Susan Dumais<\/td>\n<\/tr>\n
Tuesday, July 21, 2020<\/td>\n9:00 AM\u201312:30 PM PDT<\/td>\nSecurity and Privacy in Machine Learning<\/td>\n<\/tr>\n
<\/td>\n1:00 PM\u20132:00 PM PDT<\/td>\nPanel – Beyond Fairness: Pushing ML Frontiers for Social Equity<\/td>\n<\/tr>\n
<\/td>\n9:00 PM\u201310:30 PM PDT<\/td>\nCausality and Machine Learning (special MSR India session)<\/td>\n<\/tr>\n
Wednesday, July 22, 2020<\/td>\n9:00 AM\u201312:30 PM PDT<\/td>\nInterpretability and Explanation<\/td>\n<\/tr>\n
Thursday, July 23, 2020<\/td>\n9:00 AM\u201312:40 PM PDT<\/td>\nMachine Learning Systems (topics include NLP and Climate Impact)<\/td>\n<\/tr>\n
<\/td>\n12:40 PM\u201312:45 PM PDT<\/td>\nClosing Remarks<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
<\/div>\n
<\/div>\n

Program Committee<\/h3>\n

Vani Mandava (opens in new tab)<\/span><\/a>, Sean Kuno (opens in new tab)<\/span><\/a>, Kalika Bali (opens in new tab)<\/span><\/a>, Debadeepta Dey (opens in new tab)<\/span><\/a>, Christopher Bishop (opens in new tab)<\/span><\/a>, Asli Celikyilmaz (opens in new tab)<\/span><\/a>, Adam Trischler (opens in new tab)<\/span><\/a><\/p>\n

<\/div>\n

MSR Events and Media<\/h3>\n

Sara Smith, Jen Viencek, Jeremy Crawford and RTE Media team<\/p>\n

<\/div>\n
<\/div>\n

Microsoft\u2019s Event Code of Conduct<\/h3>\n

Microsoft\u2019s mission is to empower every person and every organization on the planet to achieve more. This includes virtual events Microsoft hosts and participates in, where we seek to create a respectful, friendly, and inclusive experience for all participants. As such, we do not tolerate harassing or disrespectful behavior, messages, images, or interactions by any event participant, in any form, at any aspect of the program including business and social activities, regardless of location.<\/p>\n

We do not tolerate any behavior that is degrading to any gender, race, sexual orientation or disability, or any behavior that would violate Microsoft\u2019s Anti-Harassment and Anti-Discrimination Policy, Equal Employment Opportunity Policy, or Standards of Business Conduct (opens in new tab)<\/span><\/a>. In short, the entire experience must meet our culture standards. We encourage everyone to assist in creating a welcoming and safe environment. Please report (opens in new tab)<\/span><\/a> any concerns, harassing behavior, or suspicious or disruptive activity. Microsoft reserves the right to ask attendees to leave at any time at its sole discretion.<\/p>\n

<\/div>\n
\n\t\n\t\tReport a concern\t<\/a>\n\n\t<\/div>\n

Opens in a new tab<\/span><\/p>\n

Monday, July 20, 2020<\/h2>\n\n\n\n\n\n\n\n
Time (PDT) <\/strong><\/td>\n Session <\/strong><\/td>\n Speaker(s) <\/strong><\/td>\n<\/tr>\n
9:00 AM-9:10 AM<\/td>\nWelcome and Kick-Off<\/td>\nSandy Blyth (opens in new tab)<\/span><\/a>, Managing Director
\nMicrosoft Research Outreach<\/td>\n<\/tr>\n
9:10 AM\u201310:00 AM<\/td>\nFireside Chat
\n[Video (opens in new tab)<\/span><\/a>]<\/td>\n		Christopher Bishop (opens in new tab)<\/span><\/a>, Technical Fellow and Lab Director
\nMicrosoft Research Cambridge<\/p>\n
10:00 AM\u201310:30 AM<\/td>\nBREAK<\/td>\n<\/td>\n<\/tr>\n
10:30 AM\u201312:00 PM<\/td>\nMachine Learning Conversations
\n[Video (opens in new tab)<\/span><\/a>]<\/td>\n		Susan Dumais (opens in new tab)<\/span><\/a>, Technical Fellow & Managing Director
\nMicrosoft Research New England, New York City and Montreal<\/p>\n
<\/div>\n
<\/div>\n
<\/div>\n

Katja Hofmann (opens in new tab)<\/span><\/a>, Principal Researcher
\nMicrosoft Research Cambridge
\nLearning to Adapt: Advances in Deep Meta Reinforcement Learning<\/p>\n

<\/div>\n

Akshay Krishnamurthy (opens in new tab)<\/span><\/a>, Principal Researcher
\nMicrosoft Research NYC
\nGeneralization and Exploration in Reinforcement Learning<\/p>\n

<\/div>\n

Asli Celikyilmaz (opens in new tab)<\/span><\/a>, Principal Researcher
\nMicrosoft Research AI
\nModeling Discourse in Long-Text Generation<\/p>\n

<\/div>\n

Dan Klein (opens in new tab)<\/span><\/a>, Technical Fellow
\nMicrosoft Semantics Machine Research
\nConversational AI: A View from Semantic Machines<\/p>\n

<\/div>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
<\/div>\n

Opens in a new tab<\/span><\/p>\n

Tuesday, July 21, 2020<\/h2>\n

Theme: Security and Privacy in Machine Learning<\/h3>\n\n\n\n\n\n\n\n\n\n
Time (PDT) <\/strong><\/td>\n Session <\/strong><\/td>\n Speaker \/ Talk Title <\/strong><\/td>\n<\/tr>\n
9:00 AM\u201310:30 AM<\/td>\nAccelerating Machine Learning with Confidential Computing<\/b>
\n[Video<\/a>]<\/p>\n
<\/div>\n
<\/div>\n
<\/div>\n

Session Leads:<\/b> Alex Shamis<\/a>, Microsoft and Stavros Volos<\/a>, Microsoft<\/p>\n

<\/div>\n

Session Abstract:<\/b> In the recent years, Machine Learning (ML) has facilitated key applications, such as medical imaging, video analytics, and financial forecasting. Understanding the massive computing requirements of ML, cloud providers have been investing in accelerated ML computing and a range of ML services. A key concern in such systems, however, is the privacy of the sensitive data being analyzed and the confidentiality of the trained models. Confidential cloud computing provides a vehicle for privacy-preserving ML, enabling multiple entities to collaborate and train accurate models using sensitive data, and to serve these models with assurance that their data and models remain protected, even from privileged attackers. In this session, our speakers will demonstrate applications and advancements in Confidential ML: (i) how confidential computing hardware can accelerate multi-party and collaborative training, creating an incentive for data sharing; and (ii) how emerging cloud accelerator systems can be re-designed to deliver strong privacy guarantees, overcoming the limited performance of CPU-based confidential computing.<\/td>\n

Antoine Delignat-Lavaud<\/a>, Microsoft
\nMulti-party Machine Learning with Azure Confidential Computing<\/p>\n
10:30 AM\u201311:00 AM<\/td>\nBREAK<\/td>\n<\/td>\n<\/tr>\n
11:00 AM\u201312:30 PM<\/td>\nSecurity and Machine Learning<\/b>
\n[Video<\/a>]<\/p>\n		Aleksander M\u0105dry (opens in new tab)<\/span><\/a>, Massachusetts Institute of Technology
\nWhat Do Our Models Learn?<\/p>\n
12:30 PM\u20131:00 PM<\/td>\nBREAK<\/td>\n<\/td>\n<\/tr>\n
1:00 PM\u20132:00 PM<\/td>\nPanel – Beyond Fairness: Pushing ML Frontiers for Social Equity<\/b>
\n[Video<\/a>]<\/p>\n
<\/div>\n
<\/div>\n
<\/div>\n

Moderator:<\/b> Mary Gray<\/a>, Microsoft<\/p>\n

<\/div>\n

Session Abstract:<\/b> At its core, machine learning is the artful science of statistically divining patterns from stores of data\u2014typically, lots of data. Much of these data are drawn from sources as diverse as tweets and Creative Commons images to COVID-19 patient health records. Machine learning uses innovative techniques to draw what it can from the data on hand to push the boundaries of such problems as reliability and robustness in algorithmic modeling; theories and applications of causal inference; development of stable, predictive models from sparse data; uses of interpretable machine learning for course-correcting models that confound reason; and finding new ways to use noisy or sparse annotated training data to drive insights. While societal impact and social equity are relevant to the frontiers above, this panel asks: How might ML take up data and questions across a variety of domains such as education, development, discrimination, housing, health disparities, inequality in labor markets, to advance our understanding of systemic inequities and challenges? These systems, arguably, tacitly shape the data, theory, and methods core to ML. How might centering questions of social equity advance the frontiers of the field?<\/td>\n

Rediet Abebe (opens in new tab)<\/span><\/a>, University of California, Berkeley<\/p>\n
9:00 PM\u201310:30 PM<\/p>\n
<\/div>\n

(9:30 AM – 11:00 AM IST
\nWednesday)<\/td>\n

Big Ideas in Causality and Machine Learning<\/b>
\n[Video<\/a>]<\/p>\n		Special MSR India session<\/i><\/p>\n
<\/div>\n
<\/div>\n
<\/div>\n

Susan Athey (opens in new tab)<\/span><\/a>, Stanford University
\nCausal Inference, Consumer Choice, and the Value of Data<\/p>\n

<\/div>\n

Elias Bareinboim (opens in new tab)<\/span><\/a>, Columbia University
\nOn the Causal Foundations of Artificial Intelligence (Explainability & Decision-Making)<\/p>\n

<\/div>\n

Cheng Zhang<\/a>, Microsoft
\nA causal view on Robustness of Neural Networks<\/p>\n

<\/div>\n

Q&A panel with all 3 speakers<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

<\/div>\n

Opens in a new tab<\/span><\/p>\n

Wednesday, July 22, 2020<\/h2>\n

Theme: Interpretability and Explanation<\/h3>\n\n\n\n\n\n\n
Time (PDT) <\/strong><\/td>\n Session Title <\/strong><\/td>\n Speaker \/ Talk Title <\/strong><\/td>\n<\/tr>\n
9:00 AM\u201310:30 AM<\/td>\nMachine Learning Reliability and Robustness<\/b>
\n[Video (opens in new tab)<\/span><\/a>]<\/p>\n
<\/div>\n
<\/div>\n
<\/div>\n

Session Lead:<\/b> Besmira Nushi (opens in new tab)<\/span><\/a>, Microsoft<\/p>\n

<\/div>\n

Session Abstract:<\/b> As Machine Learning (ML) systems are increasingly becoming part of user-facing applications, their reliability and robustness are key to building and maintaining trust with users and customers, especially for high-stake domains. While advances in learning are continuously improving model performance on expectation, there is an emergent need for identifying, understanding, and mitigating cases where models may fail in unexpected ways. This session is going to discuss ML reliability and robustness from both a theoretical and empirical perspective. In particular, the session will aim at summarizing important ongoing work that focuses on reliability guarantees but also on how such guarantees translate (or not) to real-world applications. Further, the talks and the panel will aim at discussing (1) properties of ML algorithms that make them more preferable than others from a reliability and robustness lens such as interpretability, consistency, transportability etc. and (2) tooling support that is needed for ML developers to check and build for reliable and robust ML. The discussion will be grounded on real-world applications of ML in vision and language tasks, healthcare, and decision making.<\/td>\n

Thomas Dietterich (opens in new tab)<\/span><\/a>, Oregon State University
\nAnomaly Detection in Machine Learning and Computer Vision<\/p>\n
10:30 AM\u201311:00 AM<\/td>\nBREAK<\/td>\n<\/td>\n<\/tr>\n
11:00 AM\u201312:30 PM<\/td>\nSaving Lives with Interpretable ML<\/b>
\n[Video (opens in new tab)<\/span><\/a>]<\/p>\n
<\/div>\n
<\/div>\n
<\/div>\n

Session Lead:<\/b> Rich Caruana (opens in new tab)<\/span><\/a>, Microsoft<\/p>\n

<\/div>\n

Session Abstract:<\/b> This session is about Saving Lives Using Interpretable Machine Learning in HealthCare. It\u2019s critical to make sure healthcare models are safe to deploy. One challenge is that most patients are receiving treatment and that affects the data. A model might learn high blood pressure is good for you because the treatment given when you have blood pressure lowers risk compared to healthier patients with lower blood pressure. There are many ways confounding can cause models to predict crazy things. In the first presentation Rich Caruana will talk about problems that we see in healthcare data thanks to interpretable machine learning. In the second presentation, Ankur Teredesai from UW will talk about Fairness in Machine Learning for HealthCare. And in the last presentation Marzyeh Ghassemi from Toronto will talk about how Interpretable, Explainable, and Transparent AI can be Dangerous in HealthCare. Looks like an exciting lineup, so please join us!<\/td>\n

Rich Caruana (opens in new tab)<\/span><\/a>, Microsoft
\nSaving Lives with Interpretable Machine Learning<\/p>\n
<\/div>\n
<\/div>\n
<\/div>\n

Ankur Teredesai (opens in new tab)<\/span><\/a>, University of Washington
\nFairness in Healthcare AI<\/p>\n

<\/div>\n

Marzyeh Ghassemi (opens in new tab)<\/span><\/a>, University of Toronto
\nExpl-AI-n Yourself: The False Hope of Explainable Machine Learning in Healthcare<\/p>\n

<\/div>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
<\/div>\n

Opens in a new tab<\/span><\/p>\n

Thursday, July 23, 2020<\/h2>\n

Theme: Machine Learning Systems<\/h3>\n\n\n\n\n\n\n\n
Time (PDT) <\/strong><\/td>\n Session Title <\/strong><\/td>\n Speaker \/ Talk Title <\/strong><\/td>\n<\/tr>\n
9:00 AM\u201310:30 AM<\/td>\nLearning from Limited Labeled Data: Challenges and Opportunities for NLP<\/b>
\n[Video<\/a>]<\/p>\n
<\/div>\n
<\/div>\n
<\/div>\n

Session Lead:<\/b> Ahmed Hassan Awadallah<\/a>, Microsoft<\/p>\n

<\/div>\n

Session Abstract:<\/b> Modern machine learning applications have enjoyed a great boost utilizing neural networks models, allowing them to achieve state-of-the-art results on a wide range of tasks. Such models, however, require large amounts of annotated data for training. In many real-world scenarios, such data is of limited availability making it difficult to translate these gains into real-world impact. Collecting large amounts of annotated data is often difficult or even infeasible due to the time and expense of labelling data and the private and personal nature of some of these datasets. This session will discuss several approaches to address the labelled data scarcity. In particular, the session will discuss work on: (1) transfer learning techniques that can transfer knowledge between different domains or languages to reduce the need for annotated data; (2) weakly-supervised learning where distant or heuristic supervision is derived from the data itself or other available metadata; (3) and techniques which learn from user interactions or other reward signals directly with techniques such as reinforcement learning. The discussion will be grounded on real-world applications where we aspire to bring AI experiences quickly and efficiently to everyone in more tasks, markets, languages, and domains.<\/td>\n

Ahmed Hassan Awadallah<\/a>, Microsoft
\nBringing AI Experiences to Everyone<\/p>\n
10:30 AM\u201311:00 AM<\/td>\nBREAK<\/td>\n<\/td>\n<\/tr>\n
11:00 AM\u201312:40 PM<\/td>\nClimate Impact of Machine Learning<\/b>
\n[Video<\/a>]<\/p>\n
<\/div>\n
<\/div>\n
<\/div>\n

Session Lead:<\/b> Philip Rosenfield<\/a>, Microsoft<\/p>\n

<\/div>\n

Session Abstract:<\/b> Microsoft has made an ambitious commitment to remove its carbon footprint in response to the overwhelming urgency of addressing climate change. Meanwhile, recent advances in machine learning (ML) models, such as transformer-based NLP, have produced substantial gains in accuracy at the cost of exceptionally large compute resources and, correspondingly, carbon emissions from energy consumption. Understanding and mitigating the climate impact of ML has emerged at the frontier of ML research, spanning multiple areas including hardware design, computational efficiency, and incentives for carbon efficiency.<\/p>\n

The goal of this session is to identify priority areas to drive research agendas that are best-suited to efforts in academia, in industry, or in collaboration. We aim to inspire research advances and action, within both academia and industry, to improve the sustainability of machine learning hardware, software and frameworks.<\/td>\n

Nicolo Fusi<\/a>, Microsoft
\nOpening Remarks<\/p>\n
12:40 PM\u201312:45 PM<\/td>\nClosing Remarks<\/strong><\/td>\nSandy Blyth<\/a>, Managing Director
\nMicrosoft Research Outreach<\/p>\n
<\/div>\n
<\/div>\n
<\/div>\n

Vani Mandava<\/a>, Director
\nMicrosoft Research Outreach<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

<\/div>\n

Opens in a new tab<\/span><\/p>\n","protected":false},"excerpt":{"rendered":"

Contact Us: If you have questions about this event, please email us at mlevent@microsoft.com Watch on-demand Opens in a new tab The first virtual Frontiers in Machine Learning event took place from July 20-23, 2020. This four-day virtual conference brought together academics, researchers, and PhD Students. The program was rich, engaging, and filled with current […]<\/p>\n","protected":false},"featured_media":661950,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"_classifai_error":"","msr_startdate":"2020-07-20","msr_enddate":"2020-07-23","msr_location":"Virtual","msr_expirationdate":"","msr_event_recording_link":"","msr_event_link":"","msr_event_link_redirect":false,"msr_event_time":"9:00 AM\u201312:30 PM Pacific","msr_hide_region":true,"msr_private_event":false,"msr_hide_image_in_river":0,"footnotes":""},"research-area":[13556],"msr-region":[256048],"msr-event-type":[197944],"msr-video-type":[],"msr-locale":[268875],"msr-program-audience":[],"msr-post-option":[],"msr-impact-theme":[],"class_list":["post-655893","msr-event","type-msr-event","status-publish","has-post-thumbnail","hentry","msr-research-area-artificial-intelligence","msr-region-global","msr-event-type-hosted-by-microsoft","msr-locale-en_us"],"msr_about":"\n\n

Contact Us:<\/strong> If you have questions about this event, please email us at mlevent@microsoft.com<\/a><\/p>\n

\n\t\n\t\tWatch on-demand\t<\/a>\n\n\t<\/div>\n

Opens in a new tab<\/span><\/p>\n

The first virtual Frontiers in Machine Learning event took place from July 20-23, 2020.<\/p>\n

This four-day virtual conference brought together academics, researchers, and PhD Students. The program was rich, engaging, and filled with current themes and research outcomes spanning theory and practice in Machine Learning. The agenda covered talks and discussions with Microsoft researchers and academic collaborators.<\/p>\n

Agenda Overview<\/h3>\n\n\n\n\n\n\n\n\n\n\n\n
Date <\/strong><\/td>\n Time <\/strong><\/td>\n Program <\/strong><\/td>\n<\/tr>\n
Monday, July 20, 2020<\/td>\n9:00 AM\u201310:00 AM PDT<\/td>\nFireside Chat, Chris Bishop and Peter Lee<\/td>\n<\/tr>\n
<\/td>\n10:30 AM\u201312:00 PM PDT<\/td>\nMachine Learning Conversations, a panel led by Susan Dumais<\/td>\n<\/tr>\n
Tuesday, July 21, 2020<\/td>\n9:00 AM\u201312:30 PM PDT<\/td>\nSecurity and Privacy in Machine Learning<\/td>\n<\/tr>\n
<\/td>\n1:00 PM\u20132:00 PM PDT<\/td>\nPanel – Beyond Fairness: Pushing ML Frontiers for Social Equity<\/td>\n<\/tr>\n
<\/td>\n9:00 PM\u201310:30 PM PDT<\/td>\nCausality and Machine Learning (special MSR India session)<\/td>\n<\/tr>\n
Wednesday, July 22, 2020<\/td>\n9:00 AM\u201312:30 PM PDT<\/td>\nInterpretability and Explanation<\/td>\n<\/tr>\n
Thursday, July 23, 2020<\/td>\n9:00 AM\u201312:40 PM PDT<\/td>\nMachine Learning Systems (topics include NLP and Climate Impact)<\/td>\n<\/tr>\n
<\/td>\n12:40 PM\u201312:45 PM PDT<\/td>\nClosing Remarks<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
<\/div>\n
<\/div>\n

Program Committee<\/h3>\n

Vani Mandava (opens in new tab)<\/span><\/a>, Sean Kuno (opens in new tab)<\/span><\/a>, Kalika Bali (opens in new tab)<\/span><\/a>, Debadeepta Dey (opens in new tab)<\/span><\/a>, Christopher Bishop (opens in new tab)<\/span><\/a>, Asli Celikyilmaz (opens in new tab)<\/span><\/a>, Adam Trischler (opens in new tab)<\/span><\/a><\/p>\n

<\/div>\n

MSR Events and Media<\/h3>\n

Sara Smith, Jen Viencek, Jeremy Crawford and RTE Media team<\/p>\n

<\/div>\n
<\/div>\n

Microsoft\u2019s Event Code of Conduct<\/h3>\n

Microsoft\u2019s mission is to empower every person and every organization on the planet to achieve more. This includes virtual events Microsoft hosts and participates in, where we seek to create a respectful, friendly, and inclusive experience for all participants. As such, we do not tolerate harassing or disrespectful behavior, messages, images, or interactions by any event participant, in any form, at any aspect of the program including business and social activities, regardless of location.<\/p>\n

We do not tolerate any behavior that is degrading to any gender, race, sexual orientation or disability, or any behavior that would violate Microsoft\u2019s Anti-Harassment and Anti-Discrimination Policy, Equal Employment Opportunity Policy, or Standards of Business Conduct (opens in new tab)<\/span><\/a>. In short, the entire experience must meet our culture standards. We encourage everyone to assist in creating a welcoming and safe environment. Please report (opens in new tab)<\/span><\/a> any concerns, harassing behavior, or suspicious or disruptive activity. Microsoft reserves the right to ask attendees to leave at any time at its sole discretion.<\/p>\n

<\/div>\n
\n\t\n\t\tReport a concern\t<\/a>\n\n\t<\/div>\n

Opens in a new tab<\/span><\/p>\n

Monday, July 20, 2020<\/h2>\n\n\n\n\n\n\n\n
Time (PDT) <\/strong><\/td>\n Session <\/strong><\/td>\n Speaker(s) <\/strong><\/td>\n<\/tr>\n
9:00 AM-9:10 AM<\/td>\nWelcome and Kick-Off<\/td>\nSandy Blyth (opens in new tab)<\/span><\/a>, Managing Director
\nMicrosoft Research Outreach<\/td>\n<\/tr>\n
9:10 AM\u201310:00 AM<\/td>\nFireside Chat
\n[Video (opens in new tab)<\/span><\/a>]<\/td>\n		Christopher Bishop (opens in new tab)<\/span><\/a>, Technical Fellow and Lab Director
\nMicrosoft Research Cambridge<\/p>\n
10:00 AM\u201310:30 AM<\/td>\nBREAK<\/td>\n<\/td>\n<\/tr>\n
10:30 AM\u201312:00 PM<\/td>\nMachine Learning Conversations
\n[Video (opens in new tab)<\/span><\/a>]<\/td>\n		Susan Dumais (opens in new tab)<\/span><\/a>, Technical Fellow & Managing Director
\nMicrosoft Research New England, New York City and Montreal<\/p>\n
<\/div>\n
<\/div>\n
<\/div>\n

Katja Hofmann (opens in new tab)<\/span><\/a>, Principal Researcher
\nMicrosoft Research Cambridge
\nLearning to Adapt: Advances in Deep Meta Reinforcement Learning<\/p>\n

<\/div>\n

Akshay Krishnamurthy (opens in new tab)<\/span><\/a>, Principal Researcher
\nMicrosoft Research NYC
\nGeneralization and Exploration in Reinforcement Learning<\/p>\n

<\/div>\n

Asli Celikyilmaz (opens in new tab)<\/span><\/a>, Principal Researcher
\nMicrosoft Research AI
\nModeling Discourse in Long-Text Generation<\/p>\n

<\/div>\n

Dan Klein (opens in new tab)<\/span><\/a>, Technical Fellow
\nMicrosoft Semantics Machine Research
\nConversational AI: A View from Semantic Machines<\/p>\n

<\/div>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
<\/div>\n

Opens in a new tab<\/span><\/p>\n

Tuesday, July 21, 2020<\/h2>\n

Theme: Security and Privacy in Machine Learning<\/h3>\n\n\n\n\n\n\n\n\n\n
Time (PDT) <\/strong><\/td>\n Session <\/strong><\/td>\n Speaker \/ Talk Title <\/strong><\/td>\n<\/tr>\n
9:00 AM\u201310:30 AM<\/td>\nAccelerating Machine Learning with Confidential Computing<\/b>
\n[Video<\/a>]<\/p>\n
<\/div>\n
<\/div>\n
<\/div>\n

Session Leads:<\/b> Alex Shamis<\/a>, Microsoft and Stavros Volos<\/a>, Microsoft<\/p>\n

<\/div>\n

Session Abstract:<\/b> In the recent years, Machine Learning (ML) has facilitated key applications, such as medical imaging, video analytics, and financial forecasting. Understanding the massive computing requirements of ML, cloud providers have been investing in accelerated ML computing and a range of ML services. A key concern in such systems, however, is the privacy of the sensitive data being analyzed and the confidentiality of the trained models. Confidential cloud computing provides a vehicle for privacy-preserving ML, enabling multiple entities to collaborate and train accurate models using sensitive data, and to serve these models with assurance that their data and models remain protected, even from privileged attackers. In this session, our speakers will demonstrate applications and advancements in Confidential ML: (i) how confidential computing hardware can accelerate multi-party and collaborative training, creating an incentive for data sharing; and (ii) how emerging cloud accelerator systems can be re-designed to deliver strong privacy guarantees, overcoming the limited performance of CPU-based confidential computing.<\/td>\n

Antoine Delignat-Lavaud<\/a>, Microsoft
\nMulti-party Machine Learning with Azure Confidential Computing<\/p>\n
10:30 AM\u201311:00 AM<\/td>\nBREAK<\/td>\n<\/td>\n<\/tr>\n
11:00 AM\u201312:30 PM<\/td>\nSecurity and Machine Learning<\/b>
\n[Video<\/a>]<\/p>\n		Aleksander M\u0105dry<\/a>, Massachusetts Institute of Technology
\nWhat Do Our Models Learn?<\/p>\n
12:30 PM\u20131:00 PM<\/td>\nBREAK<\/td>\n<\/td>\n<\/tr>\n
1:00 PM\u20132:00 PM<\/td>\nPanel – Beyond Fairness: Pushing ML Frontiers for Social Equity<\/b>
\n[Video<\/a>]<\/p>\n
<\/div>\n
<\/div>\n
<\/div>\n

Moderator:<\/b> Mary Gray<\/a>, Microsoft<\/p>\n

<\/div>\n

Session Abstract:<\/b> At its core, machine learning is the artful science of statistically divining patterns from stores of data\u2014typically, lots of data. Much of these data are drawn from sources as diverse as tweets and Creative Commons images to COVID-19 patient health records. Machine learning uses innovative techniques to draw what it can from the data on hand to push the boundaries of such problems as reliability and robustness in algorithmic modeling; theories and applications of causal inference; development of stable, predictive models from sparse data; uses of interpretable machine learning for course-correcting models that confound reason; and finding new ways to use noisy or sparse annotated training data to drive insights. While societal impact and social equity are relevant to the frontiers above, this panel asks: How might ML take up data and questions across a variety of domains such as education, development, discrimination, housing, health disparities, inequality in labor markets, to advance our understanding of systemic inequities and challenges? These systems, arguably, tacitly shape the data, theory, and methods core to ML. How might centering questions of social equity advance the frontiers of the field?<\/td>\n

Rediet Abebe<\/a>, University of California, Berkeley<\/p>\n
9:00 PM\u201310:30 PM<\/p>\n
<\/div>\n

(9:30 AM – 11:00 AM IST
\nWednesday)<\/td>\n

Big Ideas in Causality and Machine Learning<\/b>
\n[Video<\/a>]<\/p>\n		Special MSR India session<\/i><\/p>\n
<\/div>\n
<\/div>\n
<\/div>\n

Susan Athey<\/a>, Stanford University
\nCausal Inference, Consumer Choice, and the Value of Data<\/p>\n

<\/div>\n

Elias Bareinboim<\/a>, Columbia University
\nOn the Causal Foundations of Artificial Intelligence (Explainability & Decision-Making)<\/p>\n

<\/div>\n

Cheng Zhang<\/a>, Microsoft
\nA causal view on Robustness of Neural Networks<\/p>\n

<\/div>\n

Q&A panel with all 3 speakers<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

<\/div>\n

Opens in a new tab<\/span><\/p>\n

Wednesday, July 22, 2020<\/h2>\n

Theme: Interpretability and Explanation<\/h3>\n\n\n\n\n\n\n
Time (PDT) <\/strong><\/td>\n Session Title <\/strong><\/td>\n Speaker \/ Talk Title <\/strong><\/td>\n<\/tr>\n
9:00 AM\u201310:30 AM<\/td>\nMachine Learning Reliability and Robustness<\/b>
\n[Video (opens in new tab)<\/span><\/a>]<\/p>\n
<\/div>\n
<\/div>\n
<\/div>\n

Session Lead:<\/b> Besmira Nushi (opens in new tab)<\/span><\/a>, Microsoft<\/p>\n

<\/div>\n

Session Abstract:<\/b> As Machine Learning (ML) systems are increasingly becoming part of user-facing applications, their reliability and robustness are key to building and maintaining trust with users and customers, especially for high-stake domains. While advances in learning are continuously improving model performance on expectation, there is an emergent need for identifying, understanding, and mitigating cases where models may fail in unexpected ways. This session is going to discuss ML reliability and robustness from both a theoretical and empirical perspective. In particular, the session will aim at summarizing important ongoing work that focuses on reliability guarantees but also on how such guarantees translate (or not) to real-world applications. Further, the talks and the panel will aim at discussing (1) properties of ML algorithms that make them more preferable than others from a reliability and robustness lens such as interpretability, consistency, transportability etc. and (2) tooling support that is needed for ML developers to check and build for reliable and robust ML. The discussion will be grounded on real-world applications of ML in vision and language tasks, healthcare, and decision making.<\/td>\n

Thomas Dietterich (opens in new tab)<\/span><\/a>, Oregon State University
\nAnomaly Detection in Machine Learning and Computer Vision<\/p>\n
10:30 AM\u201311:00 AM<\/td>\nBREAK<\/td>\n<\/td>\n<\/tr>\n
11:00 AM\u201312:30 PM<\/td>\nSaving Lives with Interpretable ML<\/b>
\n[Video (opens in new tab)<\/span><\/a>]<\/p>\n
<\/div>\n
<\/div>\n
<\/div>\n

Session Lead:<\/b> Rich Caruana (opens in new tab)<\/span><\/a>, Microsoft<\/p>\n

<\/div>\n

Session Abstract:<\/b> This session is about Saving Lives Using Interpretable Machine Learning in HealthCare. It\u2019s critical to make sure healthcare models are safe to deploy. One challenge is that most patients are receiving treatment and that affects the data. A model might learn high blood pressure is good for you because the treatment given when you have blood pressure lowers risk compared to healthier patients with lower blood pressure. There are many ways confounding can cause models to predict crazy things. In the first presentation Rich Caruana will talk about problems that we see in healthcare data thanks to interpretable machine learning. In the second presentation, Ankur Teredesai from UW will talk about Fairness in Machine Learning for HealthCare. And in the last presentation Marzyeh Ghassemi from Toronto will talk about how Interpretable, Explainable, and Transparent AI can be Dangerous in HealthCare. Looks like an exciting lineup, so please join us!<\/td>\n

Rich Caruana (opens in new tab)<\/span><\/a>, Microsoft
\nSaving Lives with Interpretable Machine Learning<\/p>\n
<\/div>\n
<\/div>\n
<\/div>\n

Ankur Teredesai (opens in new tab)<\/span><\/a>, University of Washington
\nFairness in Healthcare AI<\/p>\n

<\/div>\n

Marzyeh Ghassemi (opens in new tab)<\/span><\/a>, University of Toronto
\nExpl-AI-n Yourself: The False Hope of Explainable Machine Learning in Healthcare<\/p>\n

<\/div>\n<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n
<\/div>\n

Opens in a new tab<\/span><\/p>\n

Thursday, July 23, 2020<\/h2>\n

Theme: Machine Learning Systems<\/h3>\n\n\n\n\n\n\n\n
Time (PDT) <\/strong><\/td>\n Session Title <\/strong><\/td>\n Speaker \/ Talk Title <\/strong><\/td>\n<\/tr>\n
9:00 AM\u201310:30 AM<\/td>\nLearning from Limited Labeled Data: Challenges and Opportunities for NLP<\/b>
\n[Video<\/a>]<\/p>\n
<\/div>\n
<\/div>\n
<\/div>\n

Session Lead:<\/b> Ahmed Hassan Awadallah<\/a>, Microsoft<\/p>\n

<\/div>\n

Session Abstract:<\/b> Modern machine learning applications have enjoyed a great boost utilizing neural networks models, allowing them to achieve state-of-the-art results on a wide range of tasks. Such models, however, require large amounts of annotated data for training. In many real-world scenarios, such data is of limited availability making it difficult to translate these gains into real-world impact. Collecting large amounts of annotated data is often difficult or even infeasible due to the time and expense of labelling data and the private and personal nature of some of these datasets. This session will discuss several approaches to address the labelled data scarcity. In particular, the session will discuss work on: (1) transfer learning techniques that can transfer knowledge between different domains or languages to reduce the need for annotated data; (2) weakly-supervised learning where distant or heuristic supervision is derived from the data itself or other available metadata; (3) and techniques which learn from user interactions or other reward signals directly with techniques such as reinforcement learning. The discussion will be grounded on real-world applications where we aspire to bring AI experiences quickly and efficiently to everyone in more tasks, markets, languages, and domains.<\/td>\n

Ahmed Hassan Awadallah<\/a>, Microsoft
\nBringing AI Experiences to Everyone<\/p>\n
10:30 AM\u201311:00 AM<\/td>\nBREAK<\/td>\n<\/td>\n<\/tr>\n
11:00 AM\u201312:40 PM<\/td>\nClimate Impact of Machine Learning<\/b>
\n[Video<\/a>]<\/p>\n
<\/div>\n
<\/div>\n
<\/div>\n

Session Lead:<\/b> Philip Rosenfield<\/a>, Microsoft<\/p>\n

<\/div>\n

Session Abstract:<\/b> Microsoft has made an ambitious commitment to remove its carbon footprint in response to the overwhelming urgency of addressing climate change. Meanwhile, recent advances in machine learning (ML) models, such as transformer-based NLP, have produced substantial gains in accuracy at the cost of exceptionally large compute resources and, correspondingly, carbon emissions from energy consumption. Understanding and mitigating the climate impact of ML has emerged at the frontier of ML research, spanning multiple areas including hardware design, computational efficiency, and incentives for carbon efficiency.<\/p>\n

The goal of this session is to identify priority areas to drive research agendas that are best-suited to efforts in academia, in industry, or in collaboration. We aim to inspire research advances and action, within both academia and industry, to improve the sustainability of machine learning hardware, software and frameworks.<\/td>\n

Nicolo Fusi<\/a>, Microsoft
\nOpening Remarks<\/p>\n
12:40 PM\u201312:45 PM<\/td>\nClosing Remarks<\/strong><\/td>\nSandy Blyth<\/a>, Managing Director
\nMicrosoft Research Outreach<\/p>\n
<\/div>\n
<\/div>\n
<\/div>\n

Vani Mandava<\/a>, Director
\nMicrosoft Research Outreach<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

<\/div>\n

Opens in a new tab<\/span><\/p>\n","tab-content":[{"id":0,"name":"About","content":"The first virtual Frontiers in Machine Learning event took place from July 20-23, 2020.\r\n\r\nThis four-day virtual conference brought together academics, researchers, and PhD Students. The program was rich, engaging, and filled with current themes and research outcomes spanning theory and practice in Machine Learning. The agenda covered talks and discussions with Microsoft researchers and academic collaborators.\r\n

Agenda Overview<\/h3>\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n
Date <\/strong><\/td>\r\n Time <\/strong><\/td>\r\n Program <\/strong><\/td>\r\n<\/tr>\r\n
Monday, July 20, 2020<\/td>\r\n9:00 AM\u201310:00 AM PDT<\/td>\r\nFireside Chat, Chris Bishop and Peter Lee<\/td>\r\n<\/tr>\r\n
<\/td>\r\n10:30 AM\u201312:00 PM PDT<\/td>\r\nMachine Learning Conversations, a panel led by Susan Dumais<\/td>\r\n<\/tr>\r\n
Tuesday, July 21, 2020<\/td>\r\n9:00 AM\u201312:30 PM PDT<\/td>\r\nSecurity and Privacy in Machine Learning<\/td>\r\n<\/tr>\r\n
<\/td>\r\n1:00 PM\u20132:00 PM PDT<\/td>\r\nPanel - Beyond Fairness: Pushing ML Frontiers for Social Equity<\/td>\r\n<\/tr>\r\n
<\/td>\r\n9:00 PM\u201310:30 PM PDT<\/td>\r\nCausality and Machine Learning (special MSR India session)<\/td>\r\n<\/tr>\r\n
Wednesday, July 22, 2020<\/td>\r\n9:00 AM\u201312:30 PM PDT<\/td>\r\nInterpretability and Explanation<\/td>\r\n<\/tr>\r\n
Thursday, July 23, 2020<\/td>\r\n9:00 AM\u201312:40 PM PDT<\/td>\r\nMachine Learning Systems (topics include NLP and Climate Impact)<\/td>\r\n<\/tr>\r\n
<\/td>\r\n12:40 PM\u201312:45 PM PDT<\/td>\r\nClosing Remarks<\/td>\r\n<\/tr>\r\n<\/tbody>\r\n<\/table>\r\n
<\/div>\r\n
<\/div>\r\n

Program Committee<\/h3>\r\nVani Mandava<\/a>, Sean Kuno<\/a>, Kalika Bali<\/a>, Debadeepta Dey<\/a>, Christopher Bishop<\/a>, Asli Celikyilmaz<\/a>, Adam Trischler<\/a>\r\n
<\/div>\r\n

MSR Events and Media<\/h3>\r\nSara Smith, Jen Viencek, Jeremy Crawford and RTE Media team\r\n
<\/div>\r\n
<\/div>\r\n

Microsoft\u2019s Event Code of Conduct<\/h3>\r\nMicrosoft\u2019s mission is to empower every person and every organization on the planet to achieve more. This includes virtual events Microsoft hosts and participates in, where we seek to create a respectful, friendly, and inclusive experience for all participants. As such, we do not tolerate harassing or disrespectful behavior, messages, images, or interactions by any event participant, in any form, at any aspect of the program including business and social activities, regardless of location.\r\n\r\nWe do not tolerate any behavior that is degrading to any gender, race, sexual orientation or disability, or any behavior that would violate Microsoft\u2019s Anti-Harassment and Anti-Discrimination Policy, Equal Employment Opportunity Policy, or Standards of Business Conduct<\/a>. In short, the entire experience must meet our culture standards. We encourage everyone to assist in creating a welcoming and safe environment. Please report<\/a> any concerns, harassing behavior, or suspicious or disruptive activity. Microsoft reserves the right to ask attendees to leave at any time at its sole discretion.\r\n
<\/div>\r\n
[msr-button text=\"Report a concern\" url=\"https:\/\/app.convercent.com\/en-us\/Anonymous\/IssueIntake\/LandingPage\/65d3b907-0933-e611-8105-000d3ab03673\" new-window=\"true\" ]<\/div>"},{"id":1,"name":"Monday, July 20","content":"

Monday, July 20, 2020<\/h2>\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n
Time (PDT) <\/strong><\/td>\r\n Session <\/strong><\/td>\r\n Speaker(s) <\/strong><\/td>\r\n<\/tr>\r\n
9:00 AM-9:10 AM<\/td>\r\nWelcome and Kick-Off<\/td>\r\nSandy Blyth<\/a>, Managing Director\r\nMicrosoft Research Outreach<\/td>\r\n<\/tr>\r\n
9:10 AM\u201310:00 AM<\/td>\r\nFireside Chat\r\n[Video<\/a>]<\/td>\r\nChristopher Bishop<\/a>, Technical Fellow and Lab Director\r\nMicrosoft Research Cambridge\r\n
10:00 AM\u201310:30 AM<\/td>\r\nBREAK<\/td>\r\n<\/td>\r\n<\/tr>\r\n
10:30 AM\u201312:00 PM<\/td>\r\nMachine Learning Conversations\r\n[Video<\/a>]<\/td>\r\nSusan Dumais<\/a>, Technical Fellow & Managing Director\r\nMicrosoft Research New England, New York City and Montreal\r\n
<\/div>\r\n
<\/div>\r\n
<\/div>\r\nKatja Hofmann<\/a>, Principal Researcher\r\nMicrosoft Research Cambridge\r\nLearning to Adapt: Advances in Deep Meta Reinforcement Learning\r\n
<\/div>\r\nAkshay Krishnamurthy<\/a>, Principal Researcher\r\nMicrosoft Research NYC\r\nGeneralization and Exploration in Reinforcement Learning\r\n
<\/div>\r\nAsli Celikyilmaz<\/a>, Principal Researcher\r\nMicrosoft Research AI\r\nModeling Discourse in Long-Text Generation\r\n
<\/div>\r\nDan Klein<\/a>, Technical Fellow\r\nMicrosoft Semantics Machine Research\r\nConversational AI: A View from Semantic Machines\r\n
<\/div><\/td>\r\n<\/tr>\r\n<\/tbody>\r\n<\/table>\r\n
<\/div>"},{"id":2,"name":"Tuesday, July 21","content":"

Tuesday, July 21, 2020<\/h2>\r\n

Theme: Security and Privacy in Machine Learning<\/h3>\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n
Time (PDT) <\/strong><\/td>\r\n Session <\/strong><\/td>\r\n Speaker \/ Talk Title <\/strong><\/td>\r\n<\/tr>\r\n
9:00 AM\u201310:30 AM<\/td>\r\nAccelerating Machine Learning with Confidential Computing<\/b>\r\n[Video<\/a>]\r\n
<\/div>\r\n
<\/div>\r\n
<\/div>\r\nSession Leads:<\/b> Alex Shamis<\/a>, Microsoft and Stavros Volos<\/a>, Microsoft\r\n
<\/div>\r\nSession Abstract:<\/b> In the recent years, Machine Learning (ML) has facilitated key applications, such as medical imaging, video analytics, and financial forecasting. Understanding the massive computing requirements of ML, cloud providers have been investing in accelerated ML computing and a range of ML services. A key concern in such systems, however, is the privacy of the sensitive data being analyzed and the confidentiality of the trained models. Confidential cloud computing provides a vehicle for privacy-preserving ML, enabling multiple entities to collaborate and train accurate models using sensitive data, and to serve these models with assurance that their data and models remain protected, even from privileged attackers. In this session, our speakers will demonstrate applications and advancements in Confidential ML: (i) how confidential computing hardware can accelerate multi-party and collaborative training, creating an incentive for data sharing; and (ii) how emerging cloud accelerator systems can be re-designed to deliver strong privacy guarantees, overcoming the limited performance of CPU-based confidential computing.<\/td>\r\n
Antoine Delignat-Lavaud<\/a>, Microsoft\r\nMulti-party Machine Learning with Azure Confidential Computing\r\n
10:30 AM\u201311:00 AM<\/td>\r\nBREAK<\/td>\r\n<\/td>\r\n<\/tr>\r\n
11:00 AM\u201312:30 PM<\/td>\r\nSecurity and Machine Learning<\/b>\r\n[Video<\/a>]\r\nAleksander M\u0105dry<\/a>, Massachusetts Institute of Technology\r\nWhat Do Our Models Learn?\r\n
12:30 PM\u20131:00 PM<\/td>\r\nBREAK<\/td>\r\n<\/td>\r\n<\/tr>\r\n
1:00 PM\u20132:00 PM<\/td>\r\nPanel - Beyond Fairness: Pushing ML Frontiers for Social Equity<\/b>\r\n[Video<\/a>]\r\n
<\/div>\r\n
<\/div>\r\n
<\/div>\r\nModerator:<\/b> Mary Gray<\/a>, Microsoft\r\n
<\/div>\r\nSession Abstract:<\/b> At its core, machine learning is the artful science of statistically divining patterns from stores of data\u2014typically, lots of data. Much of these data are drawn from sources as diverse as tweets and Creative Commons images to COVID-19 patient health records. Machine learning uses innovative techniques to draw what it can from the data on hand to push the boundaries of such problems as reliability and robustness in algorithmic modeling; theories and applications of causal inference; development of stable, predictive models from sparse data; uses of interpretable machine learning for course-correcting models that confound reason; and finding new ways to use noisy or sparse annotated training data to drive insights. While societal impact and social equity are relevant to the frontiers above, this panel asks: How might ML take up data and questions across a variety of domains such as education, development, discrimination, housing, health disparities, inequality in labor markets, to advance our understanding of systemic inequities and challenges? These systems, arguably, tacitly shape the data, theory, and methods core to ML. How might centering questions of social equity advance the frontiers of the field?<\/td>\r\n
Rediet Abebe<\/a>, University of California, Berkeley\r\n
9:00 PM\u201310:30 PM\r\n
<\/div>\r\n(9:30 AM - 11:00 AM IST\r\nWednesday)<\/td>\r\n
Big Ideas in Causality and Machine Learning<\/b>\r\n[Video<\/a>]\r\nSpecial MSR India session<\/i>\r\n
<\/div>\r\n
<\/div>\r\n
<\/div>\r\nSusan Athey<\/a>, Stanford University\r\nCausal Inference, Consumer Choice, and the Value of Data\r\n
<\/div>\r\nElias Bareinboim<\/a>, Columbia University\r\nOn the Causal Foundations of Artificial Intelligence (Explainability & Decision-Making)\r\n
<\/div>\r\nCheng Zhang<\/a>, Microsoft\r\nA causal view on Robustness of Neural Networks\r\n
<\/div>\r\nQ&A panel with all 3 speakers<\/td>\r\n<\/tr>\r\n<\/tbody>\r\n<\/table>\r\n
<\/div>"},{"id":3,"name":"Wednesday, July 22","content":"

Wednesday, July 22, 2020<\/h2>\r\n

Theme: Interpretability and Explanation<\/h3>\r\n\r\n\r\n\r\n\r\n\r\n\r\n
Time (PDT) <\/strong><\/td>\r\n Session Title <\/strong><\/td>\r\n Speaker \/ Talk Title <\/strong><\/td>\r\n<\/tr>\r\n
9:00 AM\u201310:30 AM<\/td>\r\nMachine Learning Reliability and Robustness<\/b>\r\n[Video<\/a>]\r\n
<\/div>\r\n
<\/div>\r\n
<\/div>\r\nSession Lead:<\/b> Besmira Nushi<\/a>, Microsoft\r\n
<\/div>\r\nSession Abstract:<\/b> As Machine Learning (ML) systems are increasingly becoming part of user-facing applications, their reliability and robustness are key to building and maintaining trust with users and customers, especially for high-stake domains. While advances in learning are continuously improving model performance on expectation, there is an emergent need for identifying, understanding, and mitigating cases where models may fail in unexpected ways. This session is going to discuss ML reliability and robustness from both a theoretical and empirical perspective. In particular, the session will aim at summarizing important ongoing work that focuses on reliability guarantees but also on how such guarantees translate (or not) to real-world applications. Further, the talks and the panel will aim at discussing (1) properties of ML algorithms that make them more preferable than others from a reliability and robustness lens such as interpretability, consistency, transportability etc. and (2) tooling support that is needed for ML developers to check and build for reliable and robust ML. The discussion will be grounded on real-world applications of ML in vision and language tasks, healthcare, and decision making.<\/td>\r\n
Thomas Dietterich<\/a>, Oregon State University\r\nAnomaly Detection in Machine Learning and Computer Vision\r\n
10:30 AM\u201311:00 AM<\/td>\r\nBREAK<\/td>\r\n<\/td>\r\n<\/tr>\r\n
11:00 AM\u201312:30 PM<\/td>\r\nSaving Lives with Interpretable ML<\/b>\r\n[Video<\/a>]\r\n
<\/div>\r\n
<\/div>\r\n
<\/div>\r\nSession Lead:<\/b> Rich Caruana<\/a>, Microsoft\r\n
<\/div>\r\nSession Abstract:<\/b> This session is about Saving Lives Using Interpretable Machine Learning in HealthCare. It\u2019s critical to make sure healthcare models are safe to deploy. One challenge is that most patients are receiving treatment and that affects the data. A model might learn high blood pressure is good for you because the treatment given when you have blood pressure lowers risk compared to healthier patients with lower blood pressure. There are many ways confounding can cause models to predict crazy things. In the first presentation Rich Caruana will talk about problems that we see in healthcare data thanks to interpretable machine learning. In the second presentation, Ankur Teredesai from UW will talk about Fairness in Machine Learning for HealthCare. And in the last presentation Marzyeh Ghassemi from Toronto will talk about how Interpretable, Explainable, and Transparent AI can be Dangerous in HealthCare. Looks like an exciting lineup, so please join us!<\/td>\r\n
Rich Caruana<\/a>, Microsoft\r\nSaving Lives with Interpretable Machine Learning\r\n
<\/div>\r\n
<\/div>\r\n
<\/div>\r\nAnkur Teredesai<\/a>, University of Washington\r\nFairness in Healthcare AI\r\n
<\/div>\r\nMarzyeh Ghassemi<\/a>, University of Toronto\r\nExpl-AI-n Yourself: The False Hope of Explainable Machine Learning in Healthcare\r\n
<\/div><\/td>\r\n<\/tr>\r\n<\/tbody>\r\n<\/table>\r\n
<\/div>"},{"id":4,"name":"Thursday, July 23","content":"

Thursday, July 23, 2020<\/h2>\r\n

Theme: Machine Learning Systems<\/h3>\r\n\r\n\r\n\r\n\r\n\r\n\r\n\r\n
Time (PDT) <\/strong><\/td>\r\n Session Title <\/strong><\/td>\r\n Speaker \/ Talk Title <\/strong><\/td>\r\n<\/tr>\r\n
9:00 AM\u201310:30 AM<\/td>\r\nLearning from Limited Labeled Data: Challenges and Opportunities for NLP<\/b>\r\n[Video<\/a>]\r\n
<\/div>\r\n
<\/div>\r\n
<\/div>\r\nSession Lead:<\/b> Ahmed Hassan Awadallah<\/a>, Microsoft\r\n
<\/div>\r\nSession Abstract:<\/b> Modern machine learning applications have enjoyed a great boost utilizing neural networks models, allowing them to achieve state-of-the-art results on a wide range of tasks. Such models, however, require large amounts of annotated data for training. In many real-world scenarios, such data is of limited availability making it difficult to translate these gains into real-world impact. Collecting large amounts of annotated data is often difficult or even infeasible due to the time and expense of labelling data and the private and personal nature of some of these datasets. This session will discuss several approaches to address the labelled data scarcity. In particular, the session will discuss work on: (1) transfer learning techniques that can transfer knowledge between different domains or languages to reduce the need for annotated data; (2) weakly-supervised learning where distant or heuristic supervision is derived from the data itself or other available metadata; (3) and techniques which learn from user interactions or other reward signals directly with techniques such as reinforcement learning. The discussion will be grounded on real-world applications where we aspire to bring AI experiences quickly and efficiently to everyone in more tasks, markets, languages, and domains.<\/td>\r\n
Ahmed Hassan Awadallah<\/a>, Microsoft\r\nBringing AI Experiences to Everyone\r\n
10:30 AM\u201311:00 AM<\/td>\r\nBREAK<\/td>\r\n<\/td>\r\n<\/tr>\r\n
11:00 AM\u201312:40 PM<\/td>\r\nClimate Impact of Machine Learning<\/b>\r\n[Video<\/a>]\r\n
<\/div>\r\n
<\/div>\r\n
<\/div>\r\nSession Lead:<\/b> Philip Rosenfield<\/a>, Microsoft\r\n
<\/div>\r\nSession Abstract:<\/b> Microsoft has made an ambitious commitment to remove its carbon footprint in response to the overwhelming urgency of addressing climate change. Meanwhile, recent advances in machine learning (ML) models, such as transformer-based NLP, have produced substantial gains in accuracy at the cost of exceptionally large compute resources and, correspondingly, carbon emissions from energy consumption. Understanding and mitigating the climate impact of ML has emerged at the frontier of ML research, spanning multiple areas including hardware design, computational efficiency, and incentives for carbon efficiency.\r\n\r\nThe goal of this session is to identify priority areas to drive research agendas that are best-suited to efforts in academia, in industry, or in collaboration. We aim to inspire research advances and action, within both academia and industry, to improve the sustainability of machine learning hardware, software and frameworks.<\/td>\r\n
Nicolo Fusi<\/a>, Microsoft\r\nOpening Remarks\r\n
12:40 PM\u201312:45 PM<\/td>\r\nClosing Remarks<\/strong><\/td>\r\nSandy Blyth<\/a>, Managing Director\r\nMicrosoft Research Outreach\r\n
<\/div>\r\n
<\/div>\r\n
<\/div>\r\nVani Mandava<\/a>, Director\r\nMicrosoft Research Outreach<\/td>\r\n<\/tr>\r\n<\/tbody>\r\n<\/table>\r\n
<\/div>"}],"msr_startdate":"2020-07-20","msr_enddate":"2020-07-23","msr_event_time":"9:00 AM\u201312:30 PM Pacific","msr_location":"Virtual","msr_event_link":"","msr_event_recording_link":"","msr_startdate_formatted":"July 20, 2020","msr_register_text":"Watch now","msr_cta_link":"","msr_cta_text":"","msr_cta_bi_name":"","featured_image_thumbnail":"\"illustrations","event_excerpt":"The first virtual Frontiers in Machine Learning event took place from July 20-23, 2020. This four-day virtual conference brought together academics, researchers, and PhD Students. The program was rich, engaging, and filled with current themes and research outcomes spanning theory and practice in Machine Learning. The agenda covered talks and discussions with Microsoft researchers and academic collaborators. Agenda Overview Date Time Program Monday, July 20, 2020 9:00 AM\u201310:00 AM PDT Fireside Chat, Chris Bishop and…","msr_research_lab":[],"related-researchers":[{"type":"user_nicename","display_name":"Christopher Bishop","user_id":31452,"people_section":"Section name 0","alias":"cmbishop"}],"msr_impact_theme":[],"related-academic-programs":[],"related-groups":[],"related-projects":[],"related-opportunities":[],"related-publications":[],"related-videos":[679917,679932,679953,679962,679971,679980,679989,679998,680007,680016],"related-posts":[],"_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-event\/655893","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-event"}],"about":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/types\/msr-event"}],"version-history":[{"count":23,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-event\/655893\/revisions"}],"predecessor-version":[{"id":1146960,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-event\/655893\/revisions\/1146960"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media\/661950"}],"wp:attachment":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media?parent=655893"}],"wp:term":[{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=655893"},{"taxonomy":"msr-region","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-region?post=655893"},{"taxonomy":"msr-event-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-event-type?post=655893"},{"taxonomy":"msr-video-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-video-type?post=655893"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=655893"},{"taxonomy":"msr-program-audience","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-program-audience?post=655893"},{"taxonomy":"msr-post-option","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-post-option?post=655893"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=655893"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}